Following on from our preview of the Strix we finally unleash the results from their cage and see how the ASUS card compared to the Founders Edition.
ASUS ROG Strix GTX 1070 Ti Review
ASUS ROG Strix GTX 1070 Ti Review
- Thread starter tinytomlogan
- Start date
Following on from our preview of the Strix we finally unleash the results from their cage and see how the ASUS card compared to the Founders Edition.
ASUS ROG Strix GTX 1070 Ti Review
AngryGoldfish
Old N Gold
I feel like there's something wrong here. In multiple benchmarks (Hitman, GOW4, Ashes) the 1070Ti Strix is beating the 1080 Strix. How can that be? I think you might need to retest some of these as something is obviously weird.
AlienALX
Well-known member
Drivers maybe. When the 1080 came out it wasn't that hot in DX12. You also need to remember, the 1080 has been out for a long time now, and drivers will absolutely no doubt have improved since.
Well, unless Tom tested them all again for the review.
AngryGoldfish
Old N Gold
This is a big issue with video card benchmarks in my opinion. Instead of having the most up-to-date information for each review (within reason), we have to cross-reference the data with another review of a more up-to-date card. So in this case I'm using Guru3D's review of the 1070Ti. I've gone to the Hitman benchmark, and instead of using their supplied graphs, I provide my own. As in, I take a more recent GTX 1080 review posted by them a couple of months ago and compared the two cards with that data rather than the original 1080 review from mid-2016. At 1440p, the 1070Ti scored 100 FPS in Hitman, while the 1080 scored 114 FPS. That's more indicative of what most people should be seeing. I did the same with GOW4 and the 1080 was 13% faster than the 1070Ti. Again, a more expected and accurate result.
AlienALX
Well-known member
Yeah it's kinda sucky but I can totally understand it. First you would need to basically keep every single card you review (most go on to other reviewers) and then you would need to test every single one again and again.
I didn't expect the Ti to beat the 1080. We all knew it wouldn't really. It was just to make the higher mid range cheaper. The 1070 IMO is starting to creak and show its age now, not being any faster than the old gen high end. We needed something just that bit better.
Tis a good release by Nvidia, I just hope the prices are right.
AngryGoldfish
Old N Gold
But you don't. Have one aftermarket (or reference) of each SKU and rebench that every 4-6 months (or whenever there is a purported gain to be had with a new driver). TTL and others review dozens of different versions of the same GPU. You don't need that as they all perform so close to each other. You only need one as a more accurate frame of reference. You don't even have to publish the data until it's needed. Whenever you have a moment spare, throw in a 1060 (this is with the assumption that you're allowed to keep at least one GPU from each SKU) and run the benchmarks again. You might even only need to do older games like Ashes and Hitman because of DX12, while Ghost Recon is a newer title and still uses DX11. I imagine initial drivers were quite good for the game.
Greenback
Active member
When you consider Tom says it takes about an hour per test and just looking there would be 10 cards depending how far back, it would need to be an aftermarket considering AMD's reference are normally the worse cards, and if Tom did Asus strix he already gets enough hate and accusations of Asus payroll, and when they top the graph because the latest test was on the newest drivers. ( I say Asus because they are the ones who send Tom everything they have new).
Though I do find something like this would be interesting for older gen cards to see if they improved or like Barnsley found performance had fallen with the newer drivers
NeverBackDown
AMD Enthusiast
Don't need every card retested. Just a few. 1070/70ti/80/ Vega 56/64. Nothing else is really relevant
Greenback
Active member
Though I agree to a point, I'm looking at someone coming looking for a card and the gap between a 1060 looking so big it makes it look obsolete
NeverBackDown
AMD Enthusiast
You can always research from older reviews from a lower end card to compare. Yeah drivers would be older but it won't be that dramatic. I mean you could always look for the newest reviewed 1060 and compare it for the most accurate review. It's worst case scenario but even if the driver improved 5% since an older review a 5% change in performance shouldn't be the deciding factor into an upgrade imo when looking for an upgrade. Now when choosing between a 1070 or Vega 56 it's different, but coming from for example a 970 to something above a 1060 5% won't matter
Greenback
Active member
Looking at the prices of these I'd have to wonder if it makes the 1070 obsolete on a like for like there's not enough difference to warrant not getting the TI, also nvidia putting a no OC I wonder how much difference between the £419 asus turbo and £499 strix is it enough to warrant £80
NeverBackDown
AMD Enthusiast
Yeah most people who wanted a 1080 already bought before the mining craze screwed up the market. So they lucked out. I pretty much payed the same amount for a 1070ti when I bought my 1080. No regrets
AngryGoldfish
Old N Gold
I paid around £510 for my 11Ghz memory high-end 1080 plus a free game. £470 for a card that comes stock with 8Ghz memory is not bad, but my 1080 is still perfectly valid based on the superior memory alone. I don't think it's worth it for the really expensive models (I saw some 1080s in the same price range as a 1080Ti) or if you're at 1080p, but even at £510 I think it's perfectly good value, especially for 1440p.
You can always overclock a reference card to the same levels as a non-reference card. The fans might be louder, but it's just for benchmarking purposes and a more accurate frame of FPS.
You're comparing a Ti version to a non-Ti version.....The Ti versions are always faster.
Takes a day PER test suite so if I test OC too thats another day
If we retested cards the graphs would be tiny and people would moan "why isnt that card in the graph" etc
The 1070 Ti results will probably need redoing in a month when new optimisations come out etc. Its a never ending battle
If we retested cards the graphs would be tiny and people would moan "why isnt that card in the graph" etc
The 1070 Ti results will probably need redoing in a month when new optimisations come out etc. Its a never ending battle
AngryGoldfish
Old N Gold
No, I'm comparing the 1070Ti to the 1080, which has faster memory, more cuda cores, and higher clock speeds. I'm not comparing the 1070Ti to the 1070.
Yeah, life is a never ending battle. Until it does end of course.
NeverBackDown
AMD Enthusiast
Personally I'd prefer that over what it is now. More up to date and accurate information over a massive list that has incomparable data because it was tested a year ago. I mean it's still valid but it's worse case scenario.
